SMART VIRTUAL MOUSE SYSTEM USING HAND GESTURE AND FINGERTIP DETECTION

Abstract

The rapid advancement of Human-Computer Interaction (HCI) has led to innovative input systems that replace traditional devices like the mouse and keyboard. This research investigates how hand gestures can be used to replace physical mouse inputs in a computer system, utilizing computer vision and deep learning techniques. We developed a virtual mouse system that captures hand gestures using a standard webcam, which is then processed by computer vision algorithms to control mouse functions such as cursor movement, clicking, scrolling, and zooming.

The system was tested under various conditions, including variations in hand size, gesture speed, and environmental lighting. Key findings indicate that the system achieves 99% accuracy under optimal conditions but drops to 92% in low or overexposed lighting and at distances greater than 30 cm. Furthermore, recognition of complex gestures like right-click and scrolling remains a challenge. To address these issues, future iterations will incorporate adaptive lighting correction algorithms, additional training data for gesture refinement, and alternative detection methods. Despite these challenges, the system offers a cost-effective, innovative solution for individuals with physical disabilities and presents promising potential for immersive, hands-free interaction with digital systems in the future.